Alright, so I’m still relatively new at working with Musicbrainz, but one thing I’ve noticed in regards to classical music is that we don’t seem to have a standardized way to name classical music works. Obviously, with classical music, a lot of works share similar names (there are quite a few “Symphony No. 5”, for instance), but in this case I’ve found it a bit confusing that Musicbrainz will sometimes use the original language of a work for the title (such as Schumann’s “5 Stücke im Volkston für Klavier und Cello, op. 102”), but then, for another work, use the English translation of the original language (such as Schumann’s “Six Etudes in Canonic Form, op. 56”). This leads to an odd bit of confusion, especially when trying to link works together with recordings and so on, as there’s no set way to search for a work title, and you basically have to guess which language it will be. Also, it seems to switch if they use the digits for a number (“5 Lieder”), or the word (“Five Lieder”).
Other websites and databases, such as Wikipedia and IMSLP, both use a sort of standardization to make it easier, and I was just wondering if there has been any prior discussion on this topic. I wanna help make MusicBrainz as effective as humanly possible.
4 Likes
The ultimate goal is to have them entered in their original language. But it’s not always possible to determine this. Especially for composers who were active in different places around the world, might have original titles in different languages.
Also lesser known composers might not or barely be documented on the internet, sometimes the only title we have is the one found on a release.
So it’s up to doing your best effort. Language can be updated later on (although it is a job in case it’s a piece in multiple parts or movements.)
Any other language title can be entered as an alias. When tagged as “primary” for that language, it will show the language you are using MB in after the original title, if not identical. Good to know here: when a title is set to English as primary language, it will also show for those using MB in any other language, in case there is no primary alias in their specific language.
Important when updating the language of a work: be sure to add the existing work title in the alias section, so it doesn’t get lost.
Regarding 5 vs five (“Five Lieder” is probably wrong, unless the composer deliberately chose to mix languages, it should probably be either “Fünf Lieder” or “Five Songs”), I’m not aware of a standard, but I consider determining the original title used by the composer in the first edition (or the manuscript for unpublished works), is the ultimate goal.
As for the search for a work: it takes the aliases in account. If the English alias of a German work is present, searching for the English title will reveal the work.
I’m fairly fresh on MB too, so possibly I’m not 100% right here. Some veteran users might provide a deeper insight in the language standardisation question.
4 Likes
Welcome to MusicBrainz, @ferretfyre ! I’m glad that you are choosing to spend your energy on classical music data. As you are seeing, it is complex and difficult, but worthwhile.
I think that it is important to recognise that the “classical music” (Western European art music) tradition prefers to apply descriptive names, rather than artistic names, to compositions (or “Works” in MusicBrainz). These are my terms, but I haven’t yet found well-recognised terms.
For instance, Beethoven wrote a piece for orchestra. It has the form we call “Symphony”. It was the first such piece he wrote. It is in the key of C-minor. Someone catalogued Beethoven’s works and gave it the number 21. Thus it is referred to, in the English-language classical music tradition, by a descriptive phrase which lists these characteristics: “the Symphony no. 1 in C major, op. 21”. Nothing of this phrase is particularly invented by Beethoven.
Contrast that with Mozart, who wrote an opera, thought up a name which conveyed a sort of artistic or intentional message about the piece, and wrote it on the score. Someone later catalogued his works and give it the number 521. Thus it is referred to in the language of the original score as “Il dissoluto punito, ossia il Don Giovanni, K. 527”.
There are many descriptive names one can apply to a work. One can add in or leave out elements of the description. One can change their order. One can add in nicknames for the work independent of the composer. And the descriptive elements are in a specific language, so two names which have the same elements but in different languages will be different descriptive names of the same work.
If one wants a standard way to come up with a descriptive name for a work, how does one choose from all this variation? Which one of all the possibilities goes into the database? How can write instructions which will lead two different editors to independently come up with the same result? That is the challenge of making such a standard for MusicBrainz.
Whatever the standard is, where I expect to find it is in the Style / Classical / Works style guideline. Right now it seems to completely duck the question of what Name to apply to a classical work.
It does! 
This is mostly intentional, in that IIRC there was a discussion back when we got works about it and we couldn’t figure out what the best option was, so we ended up with a general consensus of “as long as we can tell what the work is it’s fine, we can always figure this out later”.
The general Work documentation says a work’s name should be “The canonical title of the work, expressed in the language it was originally written.” - nothing in classical style really overrides that, as we’ve seen, so really “use the title the composer gave when known, and add aliases for English and other relevant languages” does seem the best choice at the moment. Especially now that aliases are actually shown in enough places that hopefully having a German, Russian or Japanese name won’t leave most editors struggling to figure out what the work is.
Another issue which is fairly unclear is catalog numbers. Having the most used opus or catalog number be part of the work title certainly makes it a lot easier to find, even if clearly something like BWV 1234 wasn’t originally put there by Bach and it’s debatable if it’s part of the “canonical title of the work”. I’d generally suggest keeping it there but not on the aliases; admittedly though that currently causes a silly issue in that having a primary English alias without the catalog number for an English work would lead to a display like “Symphony no. 1 in B major, op. 23 (Symphony no. 1 in B major)”. I’ve just skipped adding English aliases for English titles, but that’s not super sustainable long term..
1 Like
During the classical period, it was pretty common for composers to use Italian titles for their works (we still see it being used in movement tempos). For example, Mozart’s used Italian titles for many of his famous works. In addition of German Mozart also used French, Italian, Latin and English on the titles.
Work titles from the reneissance and baroque can be extremely long. What many of us know as “The Well-Tempered Clavier” wouldn’t expect to see “Das Wohl Temperirte Clavier. Oder Praeludia, und Fugen durch alle Tone und Semitonia, sowohl Tertiam majorem oder Ut Re Mi anlangend, als auch Tertiam minorem oder Re Mi Fa betreffend. Zum Nutzen und Gebrauch der Lehrbegierigen Musicalischen Jugend, als auch derer in diesem Studio schon Vorgeübten zur besondern Ergötzung”.
It would make sense to have a separate alias type to store the “original name". The original title of a work should be stored, but using it as the primary title for identification isn’t practical. Even the Köchel catalogue does not use the original titles (or languages) for many works. It’s important to remember that titles are intended to help identify the correct work. That is the reason why most of the databases containing data about musical works don’t use the original titles.
If seeing a list of works by a composer, let’s say all the composed symphonies, you wouldn’t expect a list which have some of them entered as sinfonies, sinfonias, ouvertures, intradas, divertimentos (like Salzburg symphony) & serenades (earlier Mozart works which are now often considered symphonies).
These “original names” can cause a lot of confusion. How many users searching for “symphony” would actually select “intrada” even though it would be the correct work. People typically add a new (duplicate) work when they think the correct work doesn’t exist.
1 Like
Yeah, “canonical” is not always the same as “original” If the main catalogue for a composer uses a shorter, more legible name than the original, I think there is a good argument to say that’s the canonical name nowadays.
1 Like
I have been thinking we mirror the way IMSLP titles works, with for instance numbers written in digits (ie “5 Études” vs. “Five Études”), and using “commonly accepted” English titles (ie “Pictures at an Exhibition” versus “Tableaux et un exposition”), but that’s just me, and I wanted to get more feedback before trying to do anything major.
1 Like
There is a data model aspect to this. The database stores “Title” as a single character string. However, descriptive names are actually a collection of attributes: for example, form = Symphony, sequence_number = 1, key = C major, catalog_reference = (Opus, 21), title_by_composer = none, popular_title = none. You can imagine a world where the data model allows storing all those attributes, and then the software has a variety of algorithms for generating a title string from those attributes (e.g. in English language conventions, in German language conventions).
Having a data model where “Title” is a single character string is a lot easier for the database developer. But it means style discussions about title style are a mixture of choosing which attributes to store, and which algorithm to use for combining those attributes as a single string, and how to get different editors to converge on the same resulting string, and how to facilitate users getting the results they want from searches on that single string.
I understand the value of making that tradeoff. But here we are, dealing with the consequences.